Äîêóìåíò âçÿò èç êýøà ïîèñêîâîé ìàøèíû. Àäðåñ îðèãèíàëüíîãî äîêóìåíòà : http://www.stsci.edu/~kgordon/papers/PS_files/ERE_DISM.ps.gz Äàòà èçìåíåíèÿ: Thu Apr 24 02:49:35 2003 Äàòà èíäåêñèðîâàíèÿ: Tue May 27 07:54:16 2008 Êîäèðîâêà: Ïîèñêîâûå ñëîâà: zodiacal light

Detection of Extended Red Emisson in the Diffuse Interstellar Medium
Karl D. Gordon 1 , Adolf N. Witt, and Brian C. Friedmann
Ritter Astrophysical Research Center, The University of Toledo
Toledo, OH 43606
ABSTRACT
Extended Red Emission (ERE) has been detected in many dusty astrophysical
objects and this raises the question: Is ERE present only in discrete objects or is it
an observational feature of all dust, i.e. present in the diffuse interstellar medium?
In order to answer this question, we determined the blue and red intensities of the
radiation from the diffuse interstellar medium (ISM) and examined the red intensity
for the presence of an excess above that expected for scattered light. The diffuse ISM
blue and red intensities were obtained by subtracting the integrated star and galaxy
intensities from the blue and red measurements made by the Imaging Photopolarimeter
(IPP) aboard the Pioneer 10 and 11 spacecraft. The unique characteristic of the
Pioneer measurements is that they were taken outside the zodiacal dust cloud and,
therefore, are free from zodiacal light. The color of the diffuse ISM was found to be
redder than the Pioneer intensities. If the diffuse ISM intensities were entirely due to
scattering from dust (i.e. Diffuse Galactic Light or DGL), the color of the diffuse ISM
would be bluer than the Pioneer intensities. Finding a redder color implies the presence
of an excess red intensity. Using a model for the DGL, we found the blue diffuse
ISM intensity to be entirely attributable to the DGL. The red DGL was calculated
using the blue diffuse ISM intensities and the approximately invariant color of the
DGL calculated with the DGL model. Subtracting the calculated red DGL from the
red diffuse ISM intensities resulted in the detection of an excess red intensity with an
average value of ¸10 S 10 (V) G2V . This represents the likely detection of ERE in the
diffuse ISM since Hff emission cannot account for the strength of this excess and the
only other known emission process applicable to the diffuse ISM is ERE. Thus, ERE
appears to be a general characteristic of dust. The correlation between NHI and ERE
intensity is (1:43 \Sigma 0:31)\Theta10 \Gamma29 ergs s \Gamma1 š A \Gamma1 sr \Gamma1 H atom \Gamma1 from which the ERE
photon conversion efficiency was estimated at 10 \Sigma 3%.
1. Introduction
Extended Red Emission (ERE) is a broad (\Delta– ¸ 800 š A) emission band with a peak
wavelength between 6500 š A and 8000 š A seen in many dusty astrophysical objects (Figure 1). For
1 present address: Department of Physics & Astronomy, Louisiana State University, Baton Rouge, LA 70803

-- 2 --
many years, the Red Rectangle was the only object known to possess such an emission feature (see
Schmidt, Cohen, & Margon [1980] for an excellent Red Rectangle spectrum). This all changed
with the discovery that many reflection nebulae possessed flux levels in the R and I bands in
excess of that expected from dust scattered starlight (Witt, Schild, & Kraiman 1984; Witt &
Schild 1985, 1986). The spectroscopic confirmation of the excess flux as ERE (Witt & Schild 1988;
Witt & Malin 1989; Witt et al. 1989; Witt & Boroson 1990) proved that ERE was a feature of
many, but not all, reflection nebulae. The identification of the band as emission was strengthened
by the imaging polarimetry of NGC 7023 by Watkin, Geldhill, & Scarrott (1991), which showed a
reduction in the R and I polarization where R and I excess flux existed. This convincingly proved
that the excess flux is due to an emission feature and not changes in the dust scattering properties.
Fig. 1.--- An example of Extended Red Emission is plotted. This spectrum is for the reflection
nebula NGC 2327 and has had the underlying scattering continuum subtracted (Witt 1988).
The detection of ERE in other dusty astrophysical objects quickly followed the confirmation
of ERE in reflection nebulae. Up to the date of this paper, ERE has been detected in the Red
Rectangle (Schmidt et al. 1980), reflection nebulae (see above), a dark nebula (Mattila 1979;

-- 3 --
Chlewicki & Laureijs 1987), Galactic cirrus clouds (Guhathakurta & Tyson 1989; Guhathakurta
& Cutri 1994; Szomoru & Guhathakurta 1998), planetary nebulae (Furton & Witt 1990, 1992),
H II regions (Perrin & Sivan 1992; Sivan & Perrin 1993), a nova (Scott, Evans, & Rawlings 1994),
the halo of the galaxy M82 (Perrin, Darbon, & Sivan 1995), and the 30 Doradus nebula in the
Large Magellanic Cloud (Darbon, Perrin, & Sivan 1998).
Clues as to the identity of the material that produces ERE are contained in the above
observations. The first clue comes from the wide variety of objects that show ERE. The material
must be able to survive in radically different environments: from cold, quiescent environments
(dark nebulae and Galactic cirrus clouds) to hot, dynamic environments (reflection nebulae,
planetary nebulae, and H II regions). Second, the material most likely is carbonaceous since ERE
has been detected in carbon rich planetary nebulae but not oxygen rich planetary nebulae (Furton
& Witt 1992). The robustness of carbonaceous dust material is supported by the essentially
constant C/H values found for sight lines exhibiting different dust characteristics (Sofia et al.
1997), implying that carbon is not easily exchanged between the gas and dust phases of the ISM.
The third clue comes from the spatial distribution of ERE in reflection nebulae (Witt & Schild
1988; Witt & Malin 1989; Witt et al. 1989; Witt & Boroson 1990; Rogers, Heyer, & Dewdney
1995; Lemaire et al. 1996) and planetary nebulae (Furton & Witt 1990). The ERE is strongest
in regions where H 2 is being dissociated, leading to the conclusion that warm atomic hydrogen
increases the efficiency of ERE luminescence. So, the material that produces ERE must be robust,
contain carbon, and produce ERE more efficiently in the presence of atomic hydrogen.
A prime candidate for this material is hydrogenated amorphous carbon (HAC), which was
first proposed to explain the ERE in the Red Rectangle (Duley 1985). With the discovery of ERE
in reflection nebulae and other dusty objects, the identification of HAC with ERE has strengthened
(Duley & Williams 1988; Witt & Schild 1988; Duley & Williams 1990; Witt & Furton 1994). The
identification of HAC with ERE is strongly supported by the laboratory work of Furton & Witt
(1993). They have shown that the low or non­existent photoluminescence of previously annealed
HAC or pure amorphous carbon can be greatly enhanced by exposure to atomic hydrogen and/or
ultraviolet radiation. This corresponds to the strengthening of ERE in H 2 dissociation regions
discussed above. Other ERE producing materials have been proposed, such as filmy quenched
carbonaceous composite (Sakata et al. 1992) and C 60 (Webster 1993). For a discussion of the
similarities and differences between these materials see Papoular et al. (1996).
As ERE has been detected in a large range of dusty objects, the question arises: Is ERE
present in the diffuse interstellar medium (ISM)? If the answer is yes, this would make ERE a
general characteristic of dust. A positive answer has been claimed by Duley & Whittet (1990),
who identified the very broad structure (VBS) seen in extinction curves (e.g. van Breda & Whittet
1981) as due to ERE. Jenniskens (1994) has pointed out that ERE cannot be the cause of the
VBS for two reasons. First, spectra showing VBS have had the nearby sky subtracted, effectively
removing any emission from the diffuse ISM. Second, the VBS strength does not depend on the
size of the aperture used. Hence, ERE has not been detected in the diffuse ISM. As a result, dust

-- 4 --
models for the diffuse ISM have not used the ability to produce ERE as a constraint on possible
dust grain materials (e.g. Kim & Martin 1996; Mathis 1996; Zubko, Kre/lowski, Wegner 1996;
Dwek et al. 1997; Li & Greenberg 1997).
This investigation is aimed at determining whether ERE is present in the diffuse ISM.
Detecting ERE in the diffuse ISM is much more difficult than doing the same in a discrete
object. For a discrete object, ERE detection is done by subtracting a nearby sky spectrum from a
spectrum of the object (e.g. Witt & Boroson 1990). Subtracting a nearby sky spectrum removes
contributions to the object spectrum from the Earth's atmosphere (airglow), zodiacal light (dust
scattered sunlight), and Galactic background light (diffuse ISM, faint stars, and galaxies). This
results in a spectrum with contributions only from the object being studied, and any ERE is
directly attributable to that object. As the light from the diffuse ISM is part of the sky spectrum,
this method will not work for it. A different method is required.
Two of the strongest (and most difficult to model) sources in a sky spectrum are airglow and
zodiacal light (Toller 1981). Both of these sources can be avoided by simply taking observations
outside the atmosphere (for airglow) and the zodiacal dust cloud (for zodiacal light). Such
measurements have already been carried out by the Imaging Photopolarimeters (IPP) carried
aboard both Pioneer 10 and 11 (Pellicori et al. 1973; Weinberg et al. 1974). The IPP measured
the intensity of almost the entire sky in the blue (437 nm) and the red (644 nm). By using only
measurements taken when the Pioneer spacecraft were beyond 3.27 AU, contributions from the
zodiacal light are avoided (Hanner et al. 1974). Therefore, the only known sources contributing to
the IPP measurements are stars, galaxies, and the diffuse ISM. Using photometric star and galaxy
catalogs, the contribution to the IPP measurements from stars and galaxies can be removed. The
resulting blue and red intensities are due only to the diffuse ISM.
While the IPP measurements have given all­sky maps at two wavelengths and not a spectrum,
this is sufficient to detect ERE in the diffuse ISM. The presence of ERE in reflection nebulae
was first detected by observing that these objects had red fluxes in excess of that expected from
dust scattered starlight (Witt, Schild, & Kraiman 1984). The diffuse ISM is a gigantic reflection
nebula with the Galaxy's starlight scattered by the Galaxy's dust. Therefore, we can use the same
criterion, excess red flux, to detect ERE in the diffuse ISM. The scattered light in the diffuse ISM
is termed Diffuse Galactic Light (DGL). The DGL will have a bluer color than the integrated
starlight because scattering by dust is more efficient at shorter wavelengths. So, if the diffuse ISM
color (red/blue ratio) is as red as or redder than the integrated starlight and other sources of
excess red light can be positively excluded, ERE is present.
Section 2 describes the Pioneer IPP measurements and the construction of the blue and
red all­sky maps. The compilation of a star and galaxy photometric catalog, complete to
approximately 20th magnitude, is detailed in section 3. The detection of ERE in the diffuse ISM
is contained in section 4. Section 5 presents the properties of the ERE in the diffuse ISM. Finally,
section 6 discusses the implications of our results and summarizes our conclusions.

-- 5 --
2. Pioneer Data
One of the instruments onboard the Pioneer 10 and 11 spacecraft was the Imaging
Photopolarimeter (IPP). The primary objectives of the IPP were to produce blue and red maps of
the brightness and polarization of the zodiacal dust cloud from 1 to 5 AU, the background light
outside the zodiacal dust cloud, and Jupiter (Pellicori et al. 1973). Of these, we were concerned
with only the all­sky blue and red surface brightness maps taken outside the zodiacal dust cloud.
The IPP was a Maksutov­type f/3.4 telescope with an aperture of 2.54 cm and a detector
consisting of a Wollaston prism, multilayer filters, and two dual­channel Bendix channeltrons
(Pellicori et al. 1973; Weinberg et al. 1974). Simultaneous measurements were made of the
orthogonal components of the electric field in both the blue and red. The spectral bandpass
(half­power) was 3950--4850 š A for the blue channel and 5900--6900 š A for the red channel (Pellicori
et al. 1973). See subsection 3.1 for more information on the photometric characteristics of
the IPP. The IPP instantaneous field of view (FOV) was 2: ffi 29 \Theta 2: ffi 29 for the background light
measurements. The IPP was mounted on a movable arm and 64 measurements were taken during
a single rotation of the Pioneer spacecraft. The angle between the arm and the Pioneer spacecraft
spin axis (look angle = L) was changed in increments of 1: ffi 83 to build up a map of the sky.
The look angle ranged between 29 ffi and 170 ffi (Pellicori et al. 1973). The effective FOV of the
measurements was 2: ffi 29 \Theta (2: ffi 29 + 5: ffi 625 sin L), with a maximum of 2: ffi 29 \Theta 7: ffi 92 when the look angle
was 90 ffi and a minimum of 2: ffi 29 \Theta 3: ffi 27 when the look angle was 170 ffi . At each look angle, a 20
data roll (rotation) measurement cycle was performed with 8 rolls for the background light, 1 for a
radioisotope­activated phosphor source ( 14 C), 1 for offset and dark current levels, and 10 for data
readout (Pellicori et al. 1973; Weinberg et al. 1974; Toller 1981).
The raw IPP background sky measurements were processed to produce the Pioneer 10/11
Background Sky data set available from the National Space Science Data Center (NSSDC).
The details of the processing can be found elsewhere (Weinberg et al. 1974; Toller 1981;
Weinberg & Schuerman 1981; Schuerman, Giovane, & Weinberg 1997). A brief description
of the processing follows. First, the data were calibrated using the inflight measurements of
the radioisotope­activated phosphor source. Second, the FOV center was computed from the
spacecraft spin axis direction, the look angle, and the clock angle. Third, the contribution from
bright stars was subtracted using the stars in the Bright Star Catalogue (Hoffleit & Warren 1991)
and stars with m V ! 8 (Toller, Tanabe, & Weinberg 1987) in the Photoelectric Catalog (Blanco
et al. 1968; Ochsenbein 1974). Fourth, 37 resolved stars were used to determine the time decay
of the instrument sensitivity and corrections to the telescope pointing. The final error in the
positions of the FOVs was on the order of 0: ffi 15--0: ffi 40. The Pioneer 10 red data have abnormally
high noise, but as we are also using Pioneer 11 data, this did not adversely affect our results. The
final Pioneer data are expressed in S 10 (V) G2V units, the equivalent number of 10th magnitude (V
band) solar­type stars per square degree. See subsection 3.1 for details of this unit.
During the cruise portion of the Pioneer 10 and 11 missions, the IPP mapped the background

-- 6 --
light a number of times. In order to determine the spatial extent of the zodiacal dust cloud,
Hanner et al. (1974) examined the brightness of two different regions of the sky as seen by the
IPP when Pioneer 10 was between 2.41 and 4.82 AU. They found that the brightness of these two
regions stopped changing after Pioneer 10 passed 3.27 AU, making this distance the outermost
detectable edge of the zodiacal dust cloud. Therefore, all the measurements taken beyond 3.27 AU
are free from detectable zodiacal light and useful for this investigation.
On 5 days while Pioneer 10 was between 3.26 and 5.15 AU and on 6 days while Pioneer 11
was between 4.06 and 4.66 AU, the IPP mapped the background light in the sky. The resolution of
a map made on a single day is determined by the FOV of the IPP and its overlap with neighboring
FOVs. Figure 2a gives an example of the pattern of FOVs using the Pioneer 10 measurements from
day 68 of 1974. One of the FOVs has been shaded to show the overlap of a FOV with neighboring
FOVs. The resolution of this map is variable, with each parallelogram being a resolution element.
In order to actually achieve this theoretical resolution, an algorithm must be used to extract the
information in the overlapping regions. In fact, the resolution of the IPP measurements can be
increased significantly by using measurements made on different days. Figure 2b gives the pattern
of FOVs for the 11 days used in creating the final high­resolution maps (see below). The FOVs
from different days do not overlap exactly as the Pioneer 10 and 11 missions were launched on
different trajectories and the spin axis of each spacecraft changed direction slowly as a function of
distance from the Sun.
Fig. 2.--- The pattern of the FOVs are plotted in these two figures. The pattern of FOVs from
Pioneer 10 on day 68 of 1974 is shown in (a) with points in the center of the FOVs. One FOV
is shaded to show the regions which overlap neighboring FOVs. The pattern of FOVs from the
11 days used in constructing the final maps is shown in (b). Due to the different trajectories of
Pioneer 10 and 11 and the variable spacecraft spin axis orientation, the FOVs from different days
do not overlap exactly.

-- 7 --
2.1. Map Generation Algorithm
The algorithm used to create the final maps is similar to that by Aumann, Fowler, & Melnyk
(1990). Their algorithm is called the Maximum Correlation Method (MCM) and it was able to
improve the resolution of IRAS maps by a factor of ¸ 6:5, from ¸ 4 0 to ¸ 36 00 . Our algorithm
was similar to MCM and worked in the following manner. An initial guess at the final image
(zeroth iteration) was taken as a positive flat image with 0: ffi 25 \Theta 0: ffi 25 pixels. The next iteration
was calculated from
p k+1
ij =
/
1
N
N
X
m=1
Cm
!
p k
ij (1)
where p k
ij is the surface brightness of the kth iteration image at pixel coordinates (i,j), Cm is the
correction factor for the mth IPP measurement (the surface brightness in particular FOV) which
includes p ij , and the sum was done over the N IPP measurements which include p ij . The value of
Cm was calculated from
Cm = I m
0
@ 1
Q
Q
X
p k
ij
1
A
\Gamma1
(2)
where I m is the mth IPP measurement and the sum was done over the Q pixels which are included
in the mth IPP FOV. The error, oe k
ij , in p k
ij was calculated from
oe k
ij = 1
N
v u u u t N
X
m=1
0
@ I m \Gamma
2
4 1
Q
Q
X
p k
ij
3
5
m
1
A
2
: (3)
With each iteration, the image gives an improved match to the IPP measurements. The outcome
of this algorithm is to produce an image which describes all 11 days of the Pioneer measurements.
The 11 days of IPP background light measurements that were used in constructing the final
high­resolution maps are tabulated in Table 1. While the 11 days overall possessed usable data,
a large number of individual measurements were seen to be of poor quality. There are a number
of sources for the poor quality data: incorrect subtraction of bright stars, scattered sunlight, and
corrupt data rolls (Toller 1981). The poor quality data were removed from consideration using
four criteria. First, data contaminated with scattered sunlight (data taken within 70 ffi of the sun
for Pioneer 10 and within 45 ffi for Pioneer 11) were removed. Second, all data with negative values
were removed as these were the result of interuptions in the datastream of between the spacecraft
and ground station. Third, the data were divided into 5 ffi \Theta 5 ffi boxes and data inside each box
deviating over 3 standard deviations from the average in either their blue measurements, red
measurements, or red/blue ratio were removed. Fourth, a small number of points were removed
by visual inspection. Approximately 25% of the IPP measurements were of poor quality.
The resulting good data were used as the input for the algorithm described above to produce
the final high­resolution maps. The algorithm was iterated 10 times, until little change was
detected. The best iteration map to use depended on the region being investigated. For low

-- 8 --
Table 1. IPP Usable Days
Spacecraft year day R a
[years] [days] [AU]
Pioneer 10 1972 354 3.26
Pioneer 10 1973 149 4.22
Pioneer 10 1973 237 4.64
Pioneer 10 1973 279 4.81
Pioneer 11 1974 57 3.50
Pioneer 10 1974 68 5.15
Pioneer 11 1974 106 3.81
Pioneer 11 1974 148 4.06
Pioneer 11 1974 178 4.22
Pioneer 11 1974 236 4.51
Pioneer 11 1974 267 4.66
a Sun­spacecraft distance, R, taken
from NSSDC WWW pages.

-- 9 --
Fig. 3.--- The Aitoff projection (galactic longitude of zero in the center) of the Pioneer blue image of
the sky is displayed. The resolution of the displayed map is 0: ffi 5 by 0: ffi 5. The large hole corresponds
to the Sun's location as seen from Pioneer 10/11. The intensity units are S 10 (V) G2V .

-- 10 --
galactic latitude regions where the amplitude of real structure in the maps is much larger than the
noise amplitude, the 10th iteration gave the best map. For high galactic latitude regions where the
real structure amplitude is smaller than the noise amplitude, the 1st iteration gave the best map.
The 10th iteration map for the blue is displayed in Figure 3. The 10th iteration red map is
similar to the blue map. Typical uncertainties were 2% for the blue and 3% for the red. Figure 3
can be compared directly with the blue background as seen by the Hipparcos star mapper. A map
of this background is presented in Figure 6 of Wicenec & van Leeuwen (1995). The comparison
is good both in overall strength and morphology. From this comparison, the uniqueness of the
Pioneer maps was quite apparent as the Pioneer blue map lacks the substantial zodiacal light seen
in the Hipparcos star mapper blue map.
As we were only concerned with high latitude regions, we will use the 1st iteration map for
the rest of this paper and save the higher iterations for later work. The 1st iteration blue and red
maps are just smoother versions of the 10th iteration maps.
3. Photometric Star and Galaxy Counts
In order for this investigation to succeed, the contribution to the Pioneer blue and red
measurements from stars and galaxies fainter than m V = 6:5 was needed. We have tackled this
problem by constructing a Master Catalog from three separate catalogs, each complete in a subset
of the range between 6.5 and ¸20th magnitude. Ironically, the stars and galaxies with magnitudes
between 12 and ¸20 (Palomar O & E) have the best photometric data available due to the
existence of the Automated Plate Scanner Catalog of the Palomar Sky Survey I (APS Catalog,
Pennington et al. 1993). In the magnitude range between 9 and 15 (¸V band in the north and
¸B in the south), the Guide Star Catalog (GSC, Lasker et al. 1990; Russell et al. 1990; Jenkner
et al. 1990) provides data in only one band. For the magnitude range between 6.5 and 9.5, there
exists no good complete photometric catalog. We have used a combination of catalogs (see x3.2)
to construct a Not So Bright Star Catalog (NSBS Catalog) to give the best currently available
positions and magnitudes for stars with magnitudes between 6.5 and 9.5.
3.1. Transformations Between Photometric Systems
Underlying the construction of the Master Catalog was the transformation between the
Palomar blue & red (O & E) magnitudes and Johnson B & R magnitudes to Pioneer blue &
red (PB & PR) magnitudes. The normalized response curves, R(–), for the blue and red bands
of all three photometric systems are shown in Figure 4 (Minkowsi & Abell 1963; Lamla 1982;
Toller 1981). The Palomar blue (O) response function was computed for an airmass of 1.5 (Hayes
& Latham 1975) in order to reproduce the transformation between the Palomar and Johnson
systems used in calibrating the APS Catalog (Humphreys et al. 1991). For all 6 above bands as

-- 11 --
Fig. 4.--- The response curves are plotted for the Pioneer blue and red channels (PB & PR, Toller
1981), the Johnson B and R (Lamla 1982) and the Palomar blue and red (O & E, Minkowsi &
Abell 1963).

-- 12 --
well as the Johnson V band, the band's equivalent wavelength (– eq ), equivalent bandpass (\Delta– eq ),
zero magnitude flux (F – ), and the intensity corresponding to S 10 (V) G2V and S 10 (V) A0V units were
computed and are tabulated in Table 2. The values of – eq and \Delta– eq are obtained by
– eq =
R –R(–) d–
R
R(–) d–
(4)
and
\Delta– eq =
R R(–) d–
R max
; (5)
respectively. The flux corresponding to a magnitude of zero, F – , in each band was calculated by
summing the product of the band's response curve and a calibrated spectrum of ff Lyrae (T¨ug,
White, & Lockwood 1977). The calibrated spectrum of ff Lyrae was multiplied by 1.028 before
use to account for the fact that ff Lyrae's V magnitude is 0.03 (Hoffleit & Warren 1991). The
intensity corresponding to one S 10 (V) G2V unit and one S 10 (V) A0V unit was computed by summing
the product of each band's response curve with the spectrum, set to 10th magnitude in the V
band, of the sun (Lockwood, T¨ug, & White 1992) and ff Lyrae (T¨ug, White, & Lockwood 1977),
respectively. One S 10 (V) X unit is defined as intensity equivalent to one 10th V magnitude star
of spectral type X per square degree where X is either G2V or A0V. The intensity in mag/ut 00
corresponding to S 10 (V) G2V units in the B bands, V band, and R bands is 28.5, 27.8, and 27.5
mag/ut 00 , respectively.
The transformations from the Palomar and Johnson systems to the Pioneer system were
accomplished by means of the above band response functions (Figure 4) and zero magnitude fluxes
(Table 2) along with an observational grid of stellar spectra spanning the Hertzsprung­Russell
diagram (Silva & Cornell 1992). This grid consists of spectra covering 3510­8930 š A with a
resolution of 11 š A and includes 72 spectral types spanning spectral classes O--M and luminosity
classes I--V. Most of the spectra are for solar metallicity stars, but some are for metal­rich and
metal­poor stars. The spectra were dereddened and stars of similar spectral types were averaged
to produce the final 72 spectral type spectra (Silva & Cornell 1992).
In order to check the accuracy of our transformations, the transformation from the Johnson
system to the Palomar system was computed and compared to the same transformation as
determined by Humphreys et al. (1991). Figure 5a displays the (O \Gamma B) correction as a function of
(B \Gamma V ) which transforms the Johnson B magnitude to the corresponding Palomar O magnitude.
Figure 5b displays the (E \Gamma R) correction as a function of (V \Gamma R) which transforms the Johnson
R magnitude to the corresponding Palomar E magnitude. The agreement between the (O \Gamma B)
and (E \Gamma R) corrections derived in this paper and those of Humphreys et al. (1991), validates this
method for deriving transformations between photometric systems.
The transformation from the Palomar to the Pioneer system is displayed in Figure 6 and
the transformation from the Johnson to the Pioneer system is shown in Figure 7. We have
fitted the resulting curves with polynomial functions in order to have an analytic form for the
transformations. The number of terms in the fitted polynomial was determined by adding terms

-- 13 --
Table 2. Photometric Band Details
System Band – eq \Delta– eq F –
a S 10 (V) G2V S 10 (V) A0V
[ š A] [ š A] [ergs cm \Gamma2 s \Gamma1 š A \Gamma1 ] [ergs cm \Gamma2 s \Gamma1 š A \Gamma1 sr \Gamma1 ]
Johnson B 4467 1014 6.632\Theta10 \Gamma9 1.198\Theta10 \Gamma9 2.174\Theta10 \Gamma9
Palomar O 4249 1168 6.343\Theta10 \Gamma9 1.087\Theta10 \Gamma9 2.080\Theta10 \Gamma9
Pioneer PB 4370 826 6.997\Theta10 \Gamma9 1.192\Theta10 \Gamma9 2.294\Theta10 \Gamma9
Johnson V 5553 881 3.639\Theta10 \Gamma9 1.193\Theta10 \Gamma9 1.193\Theta10 \Gamma9
Johnson R 6926 2057 1.950\Theta10 \Gamma9 8.813\Theta10 \Gamma10 6.394\Theta10 \Gamma10
Palomar E 6412 386 2.289\Theta10 \Gamma9 9.828\Theta10 \Gamma10 7.505\Theta10 \Gamma10
Pioneer PR 6441 968 2.305\Theta10 \Gamma9 9.919\Theta10 \Gamma10 7.558\Theta10 \Gamma10
a F – is the flux corresponding to a magnitude of zero. See text for details.
Fig. 5.--- The transformation from the Johnson system to the Palomar system is plotted. The
(O \Gamma B) correction is displayed in (a) and the (E \Gamma R) correction is displayed in (b). Note that the
(O \Gamma B) and (E \Gamma R) corrections derived in this paper agree quite well with those from Humphreys
et al. (1991).

-- 14 --
Fig. 6.--- The transformation from the Palomar system to the Pioneer system is plotted. The
(PB \Gamma O) correction is displayed in (a) and the (PR \Gamma E) correction is displayed in (b). The
maximum corrections for both (PB \Gamma O) and (PR \Gamma E) are small.
Fig. 7.--- The transformation from the Johnson system to the Pioneer system is plotted. The
(PB \Gamma B) correction is displayed in (a) and the (PR \Gamma R) correction is displayed in (b). While the
maximum correction for (PB \Gamma B) is small, the maximum correction for (PR \Gamma R) is large due to
the significantly different values of – eq for the PR and R response curves.

-- 15 --
until the resulting fitted polynomial followed the general trend of the points. The maximum
corrections for (PB \Gamma O), (PR \Gamma E), (PB \Gamma B), and (PR \Gamma R) are ¸0.25, ¸0.10, ¸0.20, and ¸1.0,
respectively. The large maximum correction for (PR \Gamma R) is due to the large difference in the – eq
value between the Pioneer (6441 š A) and Johnson (6926 š A) systems.
3.2. Master Catalog Construction
Three star and galaxy catalogs were used in constructing the Master Catalog. The three
catalogs were the Not So Bright Star Catalog (NSBS Catalog, see below), the GSC, and the APS
Catalog. As the Master Catalog includes a large number of objects, we chose to construct it in
small pieces. The GSC is split into 9537 regions (Jenkner et al. 1990) and we used the same
regions for our Master Catalog.
For stars and galaxies with PB magnitudes between 13 and ¸20th magnitude, the APS
Catalog of the Palomar Sky Survey I (POSS I) was used. This ambitious survey is using POSS I
plates to produce a catalog of stars and galaxies with positions and Palomar O and E magnitudes
between 12th magnitude and the plate limit (¸20th magnitude). Due to difficulties in automated
photometry in crowded fields, the APS Catalog is limited to jbj ? 20 ffi (Pennington et al. 1993).
The APS Catalog is also limited by the lack of any POSS I plates below a declination \Gamma33 ffi
(Minkowsi & Abell 1963). The spatial extent of the APS Catalog defines the regions where we
can accurately remove the integrated contributions of stars and galaxies to the Pioneer data. The
positional accuracy is 0: 00 5 (Pennington et al. 1993). The O and E magnitudes were calibrated using
UBV photoelectric photometry as detailed in Humphreys (1991). The resulting random error in
magnitudes was 0.2 for stars and 0.3­0.5 for galaxies (Pennington et al. 1993). The star/galaxy
classification was done using a neural network (Odewahn et al. 1992, 1993). Regions surrounding
bright stars are not included in the APS Catalog due to scattered light from the bright stars.
We identified these holes by hand and stars and galaxies from a nearby region of equal size were
copied into the hole.
The Guide Star Catalog (GSC) was used to provide data for stars and galaxies with PB
magnitudes between 9.5 and 13. For each object, the GSC contains a position and a V (h–i ú 5600
š A) or a J (h–i = 4500 š A) magnitude (Lasker et al. 1990). The position error is 0: 00 2 \Gamma 0: 00 8 and
the magnitude error is ¸0.30 mag (Russell et al. 1990). As the GSC gives only one magnitude,
deriving accurate PB and PR magnitudes for each object was not possible. Since the GSC is
complete to 15th magnitude, a large number of the GSC objects also appear in the APS catalog.
This fact allowed us to statistically transform the GSC object magnitudes to the Pioneer system.
The average (PB \Gamma V ) or (PB \Gamma J) and (PB \Gamma PR) color for stars in common to both the GSC
and APS Catalog was computed and used to transform the magnitudes of the GSC objects to the
Pioneer system.
The catalog of stars between 6.5 V magnitude and 9.5 PB magnitude was constructed

-- 16 --
by combining three different existing catalogs: the Catalogue of Stellar Identifications (CSI,
Ochsenbein 1983), the Michigan Catalogue of Two­Dimensional Spectral Types for HD Stars
(Houk Catalog, Houk & Cowley 1975; Houk 1978; Houk 1982; Houk & Smith­Moore 1988), and
the UBV Photoelectric Photometry Catalogue (UBV Catalog, Mermilliod 1987; Mermilliod 1994).
This produced the Not So Bright Star Catalog (NSBS Catalog) with each star possessing the most
accurate position, spectral type, PB magnitude, and PR magnitude possible. The CSI was used as
the base for the NSBS Catalog. It is complete to mV ú 9:5 (Ochsenbein, Bischoff, & Egret 1981)
and is the combination of many other catalogs, notably the Henry Draper Catalog (HD Catalog,
Cannon & Pickering 1918­1924; Cannon 1925­1936; Cannon & Walton Mayall 1949) and the
Smithsonian Astrophysical Observatory Catalogue (SAO Catalog, SAO Staff 1966). Stars already
subtracted from the Pioneer data were excluded from the NSBS Catalog by excluding all stars in
the Bright Star Catalogue (Hoffleit & Warren 1991) and stars with m V ! 8 (Toller, Tanabe, &
Weinberg 1987) in the Photoelectric Catalog (Blanco et al. 1968; Ochsenbein 1974).
Information for each star in the NSBS Catalog was gathered in the following manner. The
star's position was taken from the CSI. The star's spectral type was taken (in order of preference)
from the Houk Catalog or the CSI. The star's PB and PR photometry was determined from the
star's Johnson B and R magnitudes which were calculated using one of the following algorithms. If
the star's V magnitude and (B \Gamma V ) color were available from the UBV Catalog, the R magnitude
was computed by using the fit given in Figure 8 to calculate the star's (V \Gamma R) color. If only the
star's V magnitude was available, then the star's (B \Gamma V ) color was computed from
(B \Gamma V ) = (B \Gamma V ) o +E(B \Gamma V ) (6)
where (B \Gamma V ) o is the star's unreddened color and E(B \Gamma V ) is the reddening due to dust. A
similar equation was used to compute the star's (V \Gamma R) color.
The unreddened (B \Gamma V ) o and (V \Gamma R) o were determined using fits between spectral type and
unreddened color computed using the Johnson response curves and the grid of stellar spectra (Silva
& Cornell 1992). Figure 9 displays the fits to the (B \Gamma V ) o and (V \Gamma R) o colors for luminosity class
V. Fits for luminosity classes III and I were similar. Ideally, all the stars in the NSBS Catalog
would have two­dimensional MK spectral types resulting in accurate unreddened colors. This is
true for stars in the HD Catalog (complete to m pg ú 9 [Roman & Warren 1985]) with declinations
less than \Gamma12 ffi (Houk Catalog) and a few other stars in the CSI with spectral types from other
sources (Ochsenbein 1983). For the stars with only one­dimensional temperature classes, the
unreddened (B \Gamma V ) o and (V \Gamma R) o colors were calculated by taking a weighted average of the
(B \Gamma V ) o and (V \Gamma R) o colors corresponding to luminosity classes V, III, and I. The weights were
the probability of finding a V, III, or I luminosity class star of the star's temperature class with the
star's V magnitude and absolute value of galactic latitude. These probabilities were determined
from the Houk Catalog.

-- 17 --
Fig. 8.--- The transformation between the Johnson (B \Gamma V ) to (V \Gamma R) color is plotted above. The
open circles were calculated by summing the product of the appropriate response curves and the
grid of stellar spectra (Silva & Cornell 1992).

-- 18 --
Fig. 9.--- The relationship between a luminosity class V star's temperature class and (a) (B \Gamma V ) and
(b) (V \Gamma R). The spectral type code (spcode) is a numerical representation of the star's temperature
class with O = 0, B = 10, A = 20, etc. and subclasses worth their values. For example, a A4
temperature class has a spcode = 24.
For each star, E(B \Gamma V ) and E(V \Gamma R) values were calculated from
E(B \Gamma V ) =
Ÿ E(B \Gamma V )
A V
–
A V = 0:33A V (7)
and
E(V \Gamma R) =
Ÿ E(V \Gamma R)
A V
–
A V = 0:17A V ; (8)
respectively (Whittet 1992). The values for E(B \Gamma V )=A V and E(V \Gamma R)=A V correspond to an
average Milky Way extinction curve, R V = 3:05 (Whittet 1992). A V is calculated using
A V = ffd (9)
where ff = 0:6 mag kpc \Gamma1 and d is the star's distance. In order to determine the value of ff given
above, we extracted all 19,746 stars in the NSBS Catalog with two­dimensional spectral types and
an observed (B \Gamma V ). Using these stars' two­dimensional spectral types, we calculated their (B \Gamma V )
colors from the algorithm described above with a range of ff values. For an ff = 0:6 mag kpc \Gamma1 ,
the calculated and observed (B \Gamma V ) colors were equal with no galactic latitude dependence of ff
necessary. The distance was computed by solving
m V \Gamma M V = 5 log
` d
10 pc
'
+ ffd (10)
where m V is the observed V magnitude and M V is the absolute magnitude appropriate for the
star's spectral type (Schmidt­Kaler 1982). Again, for stars without two­dimensional spectral
types, the computed E(B \Gamma V ) and E(V \Gamma R) values were a weighted average of color excesses

-- 19 --
Fig. 10.--- The combined star/galaxy counts for the NSBS Catalog, the GSC, and the APS Catalog
are displayed for two Master Catalog regions. A region at b = 19: ffi 6 (GSC #2947) is displayed in (a)
and a region at b = 74: ffi 6 (GSC #2526) in (b). The counts were normalized to reflect the counts for
one square degree and one magnitude wide bins. The error bars were computed assuming Poisson
statistics.
for V, III, and I luminosity classes. Finally, the star's B and R magnitudes were transformed to
Pioneer PB and PR magnitudes using the transformations given in x3.1.
We were only interested in the integrated star/galaxy intensity and chose to do the integration
over 5 ffi \Theta 5 ffi sized regions (see section 4). Since the integration was done over regions larger than
an individual Master Catalog region which contains ¸20,000 objects, the error in the integrated
intensity was dominated by systematic errors. The random error in the integrated intensity due to
random errors in the fluxes of individual objects was very small due to the large number of objects
contributing to the integrated intensity.
The only systematic error found was associated with the lack of two­dimensional spectral
types for many of the stars in the NSBS Catalog. By using the population statistics from the Houk
Catalog (see above), the magnitude of this error was greatly reduced. Using the Master Catalog
regions with spectral types from the Houk Catalog, we determined the remaining systematic error.
The integrated flux for the stars in these Master Catalog regions with PB magnitudes between
5.5 and 9.5 was determined two ways, using the stars' two­dimensional spectral types and using
only the stars' one­dimensional spectral types. On average, the integrated PB star flux using the
one­dimensional spectral types was 2:5 \Sigma 1:5% too high and the integrated PR flux was 3:3 \Sigma 2:5%
too low. Since there was a systematic zero point error, the PB and PR magnitudes for stars
without two­dimensional spectral types in the NSBS Catalog were corrected by adding 0.0273 and
­0.0349 to their PB and PR magnitudes, respectively. This left a random error of 1:5% and 2:5%
in the integrated PB and PR star flux from the NSBS Catalog, respectively.

-- 20 --
Fig. 11.--- The contribution to the total intensity from each half magnitude bin is plotted for the
same two regions as in the previous figure. The region at b = 19: ffi 6 is displayed in (a) and the region
at b = 74: ffi 6 in (b). The peak contribution comes from about 10th magnitude and the contribution
from objects with magnitudes ? 20 is negligible.
In order to show that the construction of the Master Catalog worked, we display the
star/galaxy counts for two Master Catalog regions in Figure 10. Figure 10a shows the counts for a
low latitude region and Figure 10b shows the counts for a high latitude region. These two figures
show the counts from the NSBS Catalog, the GSC, and the APS Catalog in the regions where
these catalogs are stated to be valid. The overlap of the three catalogs is good and within the
expected uncertainties. Additional confirmation of the Master Catalog comes from comparison of
our star counts with that of the SKY model (Cohen 1994, 1995). The star counts predicted from
the SKY model agree within Poisson statistics with our observed Master Catalog counts (Cohen
1997). See Toller (1981) for an excellent review of previous star count work.
Figure 11 plots the contribution to the total intensity from each half magnitude bin for the
same regions displayed in Figure 10. On average, the maximum contribution comes from ¸10th
magnitude objects. The scatter in the curve for magnitudes ! 10 is due to the relatively small
number of bright stars per Master Catalog region. The contribution from stars and galaxies with
magnitudes ? 20 is negligible and not the result of the Master Catalog incompleteness as the
curve falls smoothly at magnitudes well below 20.
4. Detection of ERE in the Diffuse ISM
The blue and red intensities of the diffuse ISM were determined by subtracting the integrated
star/galaxy light (ISGL) from the Pioneer maps. As the ISGL was computed from the Master
Catalog, the spatial extent of the diffuse ISM intensities was limited to regions of the sky covered

-- 21 --
by the APS Catalog. Currently, the APS Catalog includes approximately 400 plates from the
POSS I in a fairly random pattern across the sky. Due to the low resolution of the Pioneer
maps and the size of an individual POSS I plate (6: ffi 6, Minkowsi & Abell 1963), regions with
many adjoining plates were needed. Currently, only two regions have enough contiguous plates to
accurately derive the diffuse ISM intensities. Region 1 is located between (l; b) = (355 ffi ; 30 ffi ) and
(10 ffi ; 60 ffi ) and region 2 is located between (l; b) = (90 ffi ; \Gamma55 ffi ) and (120 ffi ; \Gamma20 ffi ). These two regions
have areas of 315 ut ffi and 820 ut ffi and contain over 110,000 and 746,000 objects, respectively.
In these two regions, maps of the blue and red intensities for the ISGL were computed using
the Master Catalog. The fluxes of the stars and galaxies in the Master Catalog were mapped onto
a grid in galactic longitude and latitude with 0: ffi 25 \Theta 0: ffi 25 sized pixels. The fluxes were converted to
S 10 (V) G2V units using the conversions given in Table 2 and solid angles (in steradians) computed
from
\Omega = (l 2 \Gamma l 1 ) (sin b 2 \Gamma sin b 1 ) (11)
where l is galactic longitude (in radians), b is galactic latitude (in radians), and the subscripts 1
and 2 refer to the minimum and maximum values, respectively. Equation 11 is valid for rectangles
in galactic longitude and latitude. The resulting ISGL map was smoothed to a resolution of 2 ffi to
match the Pioneer maps.
In computing the ISGL intensities, we did not include a weighting in accordance with the
dwell time of each star within the FOV of individual IPP measurements for the stars fainter than
m = 6.5. The corrections for the brighter stars (m ! 6.5) which were subtracted in the original
reduction of the IPP measurements (x2), did include such a correction. The reason for this
difference in treatment was based on the expected surface density of stars of different apparent
brightness. The fainter stars are sufficiently numerous that many are present in the field at any
one time, and on average, a star is leaving the field when another similar star is entering the field.
The passages of brighter stars through the field, on the other hand, are sufficiently singular events
that the details of these passages must be taken into account.
Two different cuts in galactic longitude of the Pioneer and ISGL intensities are displayed in
Figure 12. The first cut was taken from region 1 between galactic longitudes 0 ffi and 5 ffi . The second
cut was taken from region 2 between galactic longitudes 95 ffi and 100 ffi . Other cuts in regions 1
and 2 were examined and found to be similar to those displayed in Figure 12. The corresponding
diffuse ISM intensities were determined by subtracting the ISGL intensities from the Pioneer
intensities and are plotted in Figure 13. Figure 14 displays the PR/PB ratios (in S 10 (V) G2V units)
for the Pioneer, ISGL, and diffuse ISM intensities. From Figure 14, it is obvious that the diffuse
ISM is redder (larger PR/PB ratio) than either the Pioneer measurements or the ISGL. As the
scattered component of the diffuse ISM (DGL) is bluer (see x4.1) than the Pioneer measurements,
this requires that a nonscattered component is present in the red diffuse ISM intensity.

-- 22 --
Fig. 12.--- The red and blue intensities for the Pioneer measurements and integrated star/galaxy
light (ISGL) are plotted as a function of galactic latitude for cuts in galactic longitude between 0 ffi
and 5 ffi (a) and 95 ffi and 100 ffi (b). Each point corresponds to a 5 ffi \Theta 5 ffi region. The Pioneer error
bars were computed using the algorithm described in x2.1. The ISGL error bars were assumed to
be a conservative 5% (see x3.2).
Fig. 13.--- The red and blue intensities for the light from the diffuse ISM are plotted for cuts in
galactic longitude between 0 ffi and 5 ffi (a) and 95 ffi and 100 ffi (b). The diffuse ISM intensities were
computed by subtracting the ISGL from the Pioneer measurements.

-- 23 --
Fig. 14.--- Plotted is the red/blue ratio for the Pioneer measurements, the ISGL, and the diffuse
ISM. The first plot (a) displays the cut in galactic longitude between 0 ffi and 5 ffi and the second plot
(b) the cut between 95 ffi and 100 ffi .
4.1. Diffuse Galactic Light Model
In order to determine the strength of the nonscattered component of the diffuse ISM red
intensity, the scattered component in the red (DGL) must be removed. An accurate calculation
of the DGL should include the effects of multiple scattering, the cloudiness of the interstellar
medium, and the observed anisotropy of the illuminating radiation field. Such a model exists and
it is the Witt­Petersohn DGL model (WP model, Witt & Petersohn 1994; Friedmann 1996; Witt,
Friedmann, & Sasseen 1997). The WP model treats the Galaxy as a gigantic reflection nebula.
The details of the WP model and its use in the ultraviolet can be found in Friedmann (1996) and
Witt et al. (1997). Below, the salient points of the WP model and its adaptation to the optical
are described.
Application of the WP model to the present problem was as follows. We derived the radiation
field at different heights above and below the galactic plane from the Pioneer all­sky intensity
maps using the work of Mattila (1980a,b). The all­sky intensity maps were made by adding the
intensities of the bright stars which were removed during the Pioneer data reduction (Weinberg et
al. 1974) to the Pioneer all­sky maps after interpolating over the hole caused by the Sun's location
(see Section 2). The total dust optical depth along a particular line­of­sight was computed from
Ü = C \Theta NHI where the NHI values were from the Bell Laboratories H I Survey (Stark et al. 1992)
and the Parkes H I survey (Cleary, Haslam, & Heiles 1979; Heiles & Cleary 1979). The conversion
constant, C, was 7:58 \Theta 10 \Gamma22 cm 2 for PB and 4:71 \Theta 10 \Gamma22 cm 2 for PR. C was computed using the
average Galactic extinction curve (Whittet 1992) and hNHI =E(B \Gamma V )i = 4:93 \Theta 10 21 cm \Gamma2 mag \Gamma1
(Diplas & Savage 1994). The spectrum of cloud sizes and optical depths, except for scaling to the
PB and PR bands, was the same as that in Witt et al. (1997).

-- 24 --
The WP model uses Monte Carlo techniques to describe the radiative transfer through dust.
With these techniques, photons are followed through a dust distribution and their interaction with
the dust is parameterized by the dust optical depth (Ü ), albedo, and scattering phase function
asymmetry (g). The optical depth determines where the photon interacts, the albedo gives the
probability that the photon is scattered from a dust grain, and the scattering phase function gives
the angle at which the photon scatters.
In the blue (PB), reasonable values of the dust albedo and g are 0.50--0.70 and 0.60--0.80,
respectively (Fitzgerald, Stephens, & Witt 1976; Toller 1981; Witt et al. 1982; Witt, Schild, &
Kraiman 1984; Witt, Oliveri, & Schild 1990). Putting limits on the red (PR) values of the dust
albedo and g was more difficult as the objects best suited to studies of the dust albedo and g (i.e.
reflection nebulae) are just those objects with appreciable ERE (Witt, Schild, & Kraiman 1984;
Witt & Schild 1988). Fortunately, there are a few objects without detectable ERE. One such
object is the Bok globule investigated by Witt et al. (1990), who found that the dust albedo and
g values in this Bok globule are smaller by ¸30% in the red than in the blue. This albedo and g
difference between the blue and the red is similar to that predicted by dust grain models (Kim &
Martin 1996; Mathis 1996; Zubko, Kre/lowski, Wegner 1996; Li & Greenberg 1997), which predict
either roughly constant or decreasing albedo and g values when moving from the blue to the red.
Therefore to be conservative, we have adopted red albedo and g values equal to those in the blue.
This implies that the DGL intensities we computed in the red are the upper limits. The adopted
dust grain parameters for the model runs are listed in Table 3.
A detailed comparison of the diffuse ISM intensities with those predicted for the DGL by the
WP model is presented in Figures 15 & 16. The blue diffuse ISM intensities in Figure 15a fall well
within the WP model predictions, confirming that the blue light from the diffuse ISM is entirely
attributable to the DGL. This is not the case for Figure 15b where the blue diffuse ISM intensities
generally have the same galactic latitude dependence as the WP model predictions, but fall
consistently lower. The differences between the WP model predictions and the actual blue diffuse
Table 3. WP Model Runs
PB PR
Run albedo g albedo g
1 0.50 0.60 0.50 0.60
2 0.50 0.80 0.50 0.80
3 0.70 0.60 0.70 0.60
4 0.70 0.80 0.70 0.80

-- 25 --
Fig. 15.--- The blue intensities for the diffuse ISM as well as the WP model predictions are plotted
for cuts in galactic longitude between 0 ffi and 5 ffi (a) and 95 ffi and 100 ffi (b).
Fig. 16.--- The red intensities for the diffuse ISM as well as the WP model predictions are plotted
for cuts in galactic longitude between 0 ffi and 5 ffi (a) and 95 ffi and 100 ffi (b).

-- 26 --
Fig. 17.--- Plotted is the red/blue ratio for the Pioneer measurements, the ISGL, the diffuse ISM,
and all four of the WP model runs. The first plot (a) displays the cut in galactic longitude between
0 ffi and 5 ffi and the second plot (b) the cut between 95 ffi and 100 ffi .
ISM intensities are not surprising as the WP model DGL predictions apply to an average value of
hNHI =E(B \Gamma V )i, average dust grain parameters (albedo and g), and a radiation field derived from
the intensities seen at the Earth's position in the Galaxy. The radiation field seen by a dust cloud
located high above (or below) the Galactic plane could be different from what we calculated using
the Pioneer maps and the model of Mattila (1980a,b). The dust grain properties along particular
lines­of­sight are known to vary (e.g. Gordon et al. 1994 and references therein; Calzetti et al. 1995;
Witt, Friedmann, & Sasseen 1997). In addition, the Galactic oxygen abundance gradient found by
Smartt & Rolleston (1997) implies a dependence in the dust grain properties and hNHI =E(B \Gamma V )i
with galactic longitude since oxygen is a primary grain component. In fact, significant variations
in hNHI =E(B \Gamma V )i have been measured by Diplas & Savage (1994). The disagreement between
observations and model intensities seen in Figure 16b makes it clear that a Galaxy­wide study of
the DGL might reveal important evidence about the dependence of dust properties on galactic
longitude.
While the above three points limit the applicability of the WP model DGL intensity
predictions, the color of the DGL is fairly independent of these points because the DGL color is
determined almost entirely by wavelength dependence of the dust grain properties (albedo and
optical depth) between the blue and the red. The values of the dust grain properties we used in
the DGL model were derived from both observations and dust grain models. The dependence
of the DGL color on only the wavelength dependence of the dust grain properties is illustrated
in Figure 17 which plots the PR/PB ratio for the Pioneer, ISGL, diffuse ISM, and DGL model
intensities. The PR/PB ratio (color) of all four of the WP model runs is approximately unity,
with a slight gradient in galactic latitude. This figure clearly shows the presence of excess red
intensity over that expected from scattered red photons as the average observed PR/PB ratio of

-- 27 --
the diffuse ISM is approximately two.
Our results for the presence of an excess red intensity in the diffuse ISM are very similar
to the findings of Guhathakurta & Tyson (1989) and Guhathakurta & Cutri (1994) who found
that the (B \Gamma R) colors of individual IRAS cirrus clouds were 0.5--2 magnitudes redder than that
expected for scattered light. The indentification of the red color with ERE in cirrus clouds has
been confirmed by Szomoru & Guhathakurta (1998) through optical spectroscopy of cloud edges.
From Figure 17, the color of the diffuse ISM light is 0.5--4.0 times redder than the DGL. This
corresponds to a (PB \GammaP R) color 0.4--1.5 magnitudes redder than that expected for the DGL. Our
results cover a significantly larger region of the the sky (1135 ut ffi ) than the work of Guhathakurta
& Tyson (1989) and Guhathakurta & Cutri (1994) which was for individual regions up to 1 ut ffi .
4.2. ERE Intensity
In order to determine the excess red intensity strength, the contribution from the red DGL
must be subtracted from the red diffuse ISM intensity. As the WP model is not sufficiently
complex to reproduce the blue DGL exactly, we could not use the WP model predictions of the red
DGL. On the other hand, the color of the DGL will not be greatly affected by these deficiencies
(described at the end of x4.1). Therefore, we used the average DGL color predicted by the WP
model runs (see Figure 17) along with the observed blue diffuse ISM intensity to calculate the
expected intensity of the DGL in the red. This results in a predicted red DGL which accurately
reflects the variations of the dust grain properties, hNHI =E(B \Gamma V )i, and radiation field.
The red DGL intensities calculated using the above prescription were subtracted from the red
diffuse ISM intensities to yield a lower limit to the red intensity of the nonscattered component of
the diffuse ISM. These red intensities along with the Hff intensity expected from H recombination
in the diffuse ISM (Reynolds 1984) for the two cuts displayed in previous figures are plotted in
Figure 18. Clearly, Hff emission cannot explain more than a small fraction of the nonscattered
component of the red diffuse ISM intensity. This same conclusion was reached by Guhathakurta
& Tyson (1989) in their study of Galactic cirrus clouds and has recently been confirmed through
optical spectroscopy of cirrus cloud edges (Szomoru & Guhathakurta 1998). We identify the
excess red intensity with ERE, as there are no other known red emission processes in the diffuse
ISM. The ERE strength we determined is a lower limit on the true ERE strength as the red DGL
intensity was calculated from the WP model color prediction which was an upper limit on the true
DGL color.

-- 28 --
Fig. 18.--- The ERE intensities are plotted. The ERE intensities were determined by subtracting
the predicted red DGL intensities from the red diffuse ISM intensities. The expected intensity from
diffuse ISM Hff emission (Reynolds 1984) is also plotted using the conversion of 1 rayleigh = 0.26
S 10 (V) G2V . The first plot (a) displays the cut in galactic longitude between 0 ffi and 5 ffi and the
second plot (b) the cut between 95 ffi and 100 ffi .
5. Properties of ERE in the diffuse ISM
5.1. ERE Correlations
In order to check our identification of the red excess detected in the previous subsection
with ERE, we sought correlations of the ERE intensity with known tracers of dust: H I column
densities (Cleary, Haslam, & Heiles 1979; Heiles & Cleary 1979; Stark et al. 1992) and the COBE
Diffuse Infrared Background Experiment (DIRBE) 100, 140, and 240 ¯m intensities (Hauser et al.
1997; Toller 1997).
The correlations between H I column densities and ERE intensities for regions 1 and 2 are
displayed in Figure 19. The points in Figures 19a and b were taken from the entire areas of
regions 1 and 2 and include the representative cuts displayed in Section 4. The best fit line
for region 1 was y = (2:80 \Sigma 2:05) + (0:0140 \Sigma 0:0036)x. The best fit line for region 2 was
y = (2:01 \Sigma 3:41) + (0:0159 \Sigma 0:0066)x. The best fit lines were determined using the IDL function
LINFIT. The y­intercepts for both lines are consistent, within the uncertainties, with zero and the
correlations for both regions are equivalent within the uncertainties. Combining the two regions
yielded a best fit line of y = (2:67 \Sigma 1:71)+(0:0145 \Sigma 0:0032)x. This best fit line gives the relationship
between the ERE energy emitted per H atom and it is (1:43 \Sigma 0:31)\Theta10 \Gamma29 ergs s \Gamma1 š A \Gamma1 sr \Gamma1 H
atom \Gamma1 .
The correlations between the DIRBE 100, 140, and 240 ¯m and ERE intensities are plotted
in Figure 20. Unlike region 1, there was no clear relationship between the DIRBE and ERE

-- 29 --
Fig. 19.--- This figure displays the correlation between H I and ERE intensity for regions 1 (a) and
2 (b).
Fig. 20.--- This figure displays the correlation between the DIRBE 100, 140, and 240 ¯m intensities
and ERE intensity for regions 1 (a) and 2 (b).

-- 30 --
Fig. 21.--- This figure displays the correlation between the DIRBE 100, 140, and 240 ¯m intensities
and H I column densities for regions 1 (a) and 2 (b).
intensities for region 2. This was troubling as the correlations between the DIRBE and ERE
intensities for region 1 was quite good. The origin of the poor correlations for region 2 is related
to the fact that the DIRBE data were collected from a satellite in near­Earth space. While the
DIRBE data have had the majority of the zodiacal dust emission subtracted, there is a noticeable
residual left. This is illustrated in Figure 21b which shows the correlations between the DIRBE
intensities and H I column densities for region 2. The peak in the DIRBE intensities around
NHI ¸ 550 \Theta 10 18 cm \Gamma2 is due to residual zodiacal dust emission as the ecliptic plane runs through
the middle of region 2.
The ERE intensity was well correlated with dust tracers for region 1. For region 2, the ERE
strength was correlated with H I column density but not DIRBE intensities. The poor correlation
between the ERE and DIRBE intensities was explicable when the presence of residual zodiacal
dust emission was taken into account. Thus, the identification of the red nonscattered excess
intensity in the diffuse ISM with ERE was greatly strengthened. Additionally, this is the first time
in which it has been possible to directly correlate the strength of the ERE emission with known
dust tracers.
5.2. ERE Photon Efficiency
The process which produces ERE is thought to be photoluminescence from a disordered solid
like HAC. Basically, a UV or blue photon excites an electron from the valence to the conduction
band and the election relaxes back to the conduction band via a number of (mostly non­radiative)
transitions which might include one radiative transition. When this radiative transition occurs, an
ERE photon is emitted (Furton 1994). At most, one ERE photon is emitted per absorbed UV

-- 31 --
or blue photon. The photon efficiency is the percent of photons which are absorbed by the ERE
producing material which cause the emission of ERE photons.
We calculated the lower limit on the photon efficiency by assuming all the photons absorbed
by dust are absorbed by the ERE producing material. The calculation is fairly straightforward,
but required a couple of assumptions. The calculation begins with the relationship between the
ERE intensity and H I column density determined in the previous section. The number of ERE
photons emitted per H atom was calculated from
ERE photons
H atoms = 4úA\Delta– eq
/
hc
– eq
! \Gamma1
= 5:65 \Theta 10 \Gamma14 photons
s H atom (12)
where
A = 1:43 \Theta 10 \Gamma29 ergs
s š A sr H atom
;
\Delta– eq = 968 š A, and – eq = 6441 š A (Table 2). The number of ERE photons per unit Ü V is then
ERE photons
Ü V
= ERE photons
H atoms
NHI
E(B \Gamma V )
E(B \Gamma V )
A V
1:086 A V
Ü V
= 9:95 \Theta 10 7 photons
s cm 2 Ü V
(13)
where E(B \Gamma V )=A V = R \Gamma1
V = 3:05 \Gamma1 (Whittet 1992) and NHI=E(B \Gamma V ) = 4:93 \Theta 10 21 cm \Gamma2 mag \Gamma1
(Diplas & Savage 1994).
The other half of this calculation was to determine the number of photons absorbed per unit
Ü V . The photons absorbed per unit Ü V was calculated from
abs. photons
Ü V
=
Z 5500 š A
912 š A
(1 \Gamma a – ) (1 \Gamma e \GammaÜ – )
Ü V
I ISRF d–
ú
Z 5500 š A
912 š A
(1 \Gamma a – ) Ü –
Ü V
I ISRF d– (Ü – Ü 1)
= 7:41 \Theta 10 8 photons
s cm 2 Ü V
(14)
where a – is the wavelength dependent dust albedo (Gordon et al. 1994 and references therein), Ü –
is the wavelength dependent dust optical depth for R V = 3:05 (Cardelli, Clayton, & Mathis 1989),
and I ISRF is the intensity of the interstellar radiation field at the Earth in units of ergs cm \Gamma2 s \Gamma1
š A \Gamma1 (Witt & Johnson 1973; Mathis, Mezger, & Panagia 1983). This result is valid in the optically
thin limit (Ü – Ü 1) which is valid for the diffuse ISM far from the galactic plane. Using the results
calculated above for the number of photons absorbed and emitted, the ERE photon efficiency in
the diffuse ISM was 14%. This corresponds to an energy efficiency of 5%.

-- 32 --
Fig. 22.--- The energy emitted per H atom by the diffuse ISM is plotted. The ERE point is from
this work. The COBE points are from the work of Arendt et al. (1997) and Boulanger et al. (1996).
This calculation is based on two uncertain assumptions. The first was the strength of the
interstellar radiation field, I ISRF (Mathis, Mezger, & Panagia 1983). This I ISRF is likely to be too
low as the Mathis et al. (1983) I ISRF was calculated for a smooth diffuse ISM, whereas the diffuse
ISM is known to be clumpy (Elmegreen & Falgarone 1996) and the radiative transfer in a clumpy
medium is greatly different from that in a smooth medium (Witt & Gordon 1996). Additional
discussion of the appropriate I ISRF can be found in Mathis (1997).
We determined the actual level of the I ISRF by equating the energy absorbed by dust to
that emitted by dust. Except for ERE, the energy emitted by diffuse ISM dust is emitted in
the infrared and the Cosmic Background Explorer (COBE, Boggess et al. 1992) contained the
Far­Infrared Absolute Spectrophotometer (FIRAS) and DIRBE which measured the infrared
emission from dust. Using data from high latitude regions, the diffuse ISM spectrum per H atom
from 3.3 to 240 ¯m was derived by Arendt et al. (1997, as presented in Dwek et al. [1997]) and

-- 33 --
from 100 ¯mto 1 mm by Boulanger et al. (1996). In Figure 22, we have plotted the spectrum of
the diffuse ISM including the contribution from ERE as determined in this work. Integrating this
spectrum gives the emitted energy which is 5.77\Theta10 \Gamma24 ergs s \Gamma1 H atom \Gamma1 . The ERE contributes
a quite significant 3% of the total energy emitted by dust. In order to get the same amount of
energy absorbed by dust between 912 š A and 1.2 ¯m, the Mathis et al. (1993) I ISRF must be
multiplied by 1.33. Using this elevated I ISRF and given the uncertainties of all the quantities
contributing to this result, we calculated the lower limit ERE photon efficiency to be 10 \Sigma 3% and
the corresponding energy efficiency to be 4 \Sigma 1%.
The second assumption was that all the photons absorbed by dust were absorbed by the
material which produces ERE. This is unlikely to be true. Reducing the number of photons
absorbed by the ERE producing material would raise the ERE efficiency. Assuming a 100%
conversion efficiency for the ERE producing material would require that 10% of the photons
absorbed by the dust are converted to ERE photons. This makes the ERE producing material an
important component of dust grains.
Instead of a wavelength independent ERE conversion efficiency, perhaps it is a question of a
high efficiency in a narrow range of wavelengths. Such behavior could be caused by a strong dust
absorption feature such as the 2175 š A extinction bump. The percentage of the photons absorbed
by the 2175 š A bump material as compared the total number of photons absorbed by dust, was
calculated as the difference between the number of photons absorbed for dust between 912 š A and
5500 š A with and without a 2175 š A bump . The number of photons absorbed by bumpless dust was
calculated using equation 14, a R V = 3:05 Ü – curve from Cardelli et al. (1989) with the bump term
removed, and a – = 0:6. We calculated that 13% of the photons absorbed by dust are absorbed by
the 2175 š A absorption feature. This percentage is higher than the 10% ERE efficiency and, thus,
there are enough photons absorbed by the 2175 š A bump material to produce the observed ERE
intensity. We are not claiming the 2175 š A bump is the source of the ERE photons, only that this
is the kind of absorption feature needed if the ERE photons are absorbed in a narrow range of
wavelengths. It is intriguing that a known absorption feature, which has never been definitively
identified with a particular material, could be the source of the ERE photons.
6. Discussion and Conclusions
The detection of ERE in the diffuse ISM has direct consequences for all dust grain models.
With this work, ERE has been detected from H II regions to the diffuse ISM effectively spanning
the entire range of environments where dust is present. Thus, ERE is a characteristic of dust in
general. None of the current dust grain models (e.g. Kim & Martin 1996; Mathis 1996; Zubko,
Kre/lowski, Wegner 1996; Dwek et al. 1997; Li & Greenberg 1997) specifically includes a material
which can produce ERE. As ERE has been detected only in carbon rich planetary nebulae (Furton
& Witt 1992), this adds ERE to the already long list of dust features attributed to carbonaceous
materials. This further deepens the current ``carbon crisis'' (Snow & Witt 1995, 1996) -- the

-- 34 --
conflict between the small amount of carbon available for dust grains and the large number of
observed dust features attributed to carbonaceous materials. As the identification of other dust
features with carbonaceous materials is not as strong as that of ERE, the connection of some dust
features with carbonaceous materials needs to be reevaluated.
6.1. Anomalous Dust Scattering Properties instead of ERE?
A possible (but highly unlikely) explanation of the red excess observed in the diffuse ISM is
anomalous dust scattering properties in the red. If the dust albedo increased to unity (perfectly
scattering dust grains) around 6500 š A, it would almost be possible to explain the strength of the
red excess, because the red excess is approximately equal to the scattered intensity which was
computed for albedos between 0.5 and 0.7. In fact, this argument was presented by Greenstein
& Oke (1977) in their attempt to explain the excess red intensity of the Red Rectangle. They
concluded that only with most favorable geometry and a high albedo could they explain the excess.
A scattering explanation has been effectively ruled out by means of spectropolarimetric
results, which show that the polarization decreases where ERE increases. For example, this
explanation for the Red Rectangle was convincingly shown to be false by the spectropolarimetry of
Schmidt et al. (1980). Schmidt et al. (1980) observed a reduction in the polarization across the red
excess feature which proved that the excess was due to emission and not scattering. The emission
nature of the red excess in reflection nebulae has also been confirmed by the imaging polarimetry
of NGC 7023 by Watkin et al. (1991), who found that the polarization decreased in locations were
the ERE was the strongest. The two polarimetric observations discussed above are the only two
polarimetric observations which have been done on objects with ERE. This suggests that the red
albedo is approximately equal to the blue albedo as derived by Witt et al. (1990) for a Bok globule
without ERE and calculated from dust grain models (Kim & Martin 1996; Mathis 1996; Zubko,
Kre/lowski, Wegner 1996; Li & Greenberg 1997). Thus, anomalous red dust scattering properties
are extremely unlikely to account for the red excess in the diffuse ISM.
6.2. I ERE =I SCA and ERE efficiency
The photon efficiency calculated above for the ERE in the diffuse ISM is 10%. The question
arises: Is this efficiency characteristic of ERE in general or only ERE in the diffuse ISM? It should
be possible to calculate the ERE efficiency knowing the scattered intensity (I SCA ), the ERE
intensity, and the type of illuminating radiation field. From the scattered intensity and the type of
radiation field, the number of photons absorbed by dust in the UV and blue should be calculable.
In practice, the relationship between the scattered and absorbed intensity is quite complicated
due to the non­isotropic nature of scattering by dust grains (see x4.1).
Figure 23 plots the ERE intensity versus the scattered intensity for all objects with ERE and

-- 35 --
Fig. 23.--- This plot shows the ERE versus scattered intensities for various objects were the ERE
intensity has been measured. The diffuse ISM points are from this work. The L1780 dark cloud
point is from Mattila (1979). The reflection nebulae and the Red Rectangle points are from Witt
& Boroson (1990). The Orion H II region points were taken from Figures 3 & 4 of Perrin & Sivan
(1992). The dashed lines show I ERE =I SCA ratios of 1%, 5%, 20%, 100%, and 200% and are labeled
appropriately.

-- 36 --
scattered intensities available in the literature (Mattila 1979; Witt & Boroson 1990; Perrin &
Sivan 1992). This figure is similar to Figure 1 in Witt & Boroson (1990) but with a much larger
range in I ERE and I SCA values due to the inclusion of Orion H II region points, at the high end,
and diffuse ISM points, at the low end. The I ERE =I SCA ratio ranges from 0.01 to over 2.0. The
Red Rectangle, which was long thought to be unique in its ERE intensity, is seen to have normal
I ERE =I SCA ratios.
Interpreting the scatter in the I ERE =I SCA ratio involves detailed knowledge of the geometry
and type of illuminating radiation field in each object. The dust scattering geometry affects
I ERE =I SCA such that the larger the angle of scattering by dust grains, the lower I SCA for a given
radiation field, and the larger the I ERE =I SCA value. Essentially, the scattered intensity is reduced
due to the forward scattering nature of dust grains, while the ERE intensity is unaffected as the
dust emits ERE isotropically. The illuminating radiation field affects I ERE =I SCA since the I ERE
intensity is proportional to the radiation field between 912 and 5500 š A and I SCA is proportional
to the number of photons between 5500 and 8000 š A. Thus, the bluer the illuminating radiation
field, the larger the I ERE =I SCA .
The lack of low values of I ERE =I SCA in the diffuse ISM (0.05--2.0) is likely due to the fact
that the diffuse ISM observations were all at high galactic latitudes where the dust predominately
scatters photons from the Galactic disk at large angles. Similarly, the high value of I ERE =I SCA
(0.3) in the dark nebula L1780 is due to its location high above the Galactic disk (b ú 36 ffi )
and resulting large scattering angle. The bipolar geometry of the Red Rectangle, oriented
perpendicular to our line­of­sight, leads to predominately large angle scattering, again giving large
I ERE =I SCA values (0.5--1.5). The range in I ERE =I SCA (0.01--0.68) in reflection nebulae is mostly
due to the geometry and this is superbly illustrated as the highest reflection nebula I ERE =I SCA
value (0.68) is for IC 63, an externally illuminated nebula which scatters photons at ú 90 ffi (Witt
et al. 1989; Gordon et al. 1997). The real range of I ERE =I SCA in reflection nebulae is larger than
that represented by the spectroscopic detections of ERE shown in Figure 23. Figure 24 plots the
(B \Gamma V ) and (V \Gamma R) colors observed in 14 reflection nebulae by Witt & Schild (1986). As can be
seen from the location of the dashed line, the maximum I ERE =I SCA is at least 1.0. On the other
hand, the Orion H II region has high values of I ERE =I SCA (0.2--0.6) due to the extreme blueness
of the illuminating radiation field which originates from the hot Trapezium stars (late O, early
B spectral types). Thus, the scatter in I ERE =I SCA could easily be explained by geometry and
radiation field effects with a constant ERE efficiency of 10%. The test of this hypothesis awaits
a detailed calculation of the ERE efficiency in each object properly accounting for the effects of
geometry and illuminating radiation field.
6.3. Conclusions
Using blue and red all­sky measurements taken by Pioneer 10 and 11 outside the zodiacal
dust cloud along with star and galaxy counts to 20th magnitude, we determined the blue and red

-- 37 --
Fig. 24.--- The \Delta(B \Gamma V ) and \Delta(B \Gamma R) colors of reflection nebulae observed by Witt & Schild
(1986) are plotted. The \Delta(B \Gamma V ) color is the nebular (B \Gamma V ) color minus the illuminating star's
(B \Gamma V ). The \Delta(B \Gamma R) color is defined similarly. The solid line shows the expected relationship
between \Delta(B \Gamma V ) and \Delta(B \Gamma R) for scattering assuming the dust albedo is equal at B, V, and R
wavelengths. The dashed line gives the relationship for I ERE =I SCA = 1.

-- 38 --
intensity of the diffuse ISM in two large regions with areas of 315 ut ffi and 820 ut ffi . By comparison
with a model for the DGL, the blue diffuse ISM intensity was found to be entirely attributable to
DGL. The diffuse ISM red intensity was found to consist of DGL and ERE in roughly equal parts.
Thus, the ERE is detected in the diffuse ISM and shown to be a general characteristic of dust in
all dusty environments. The ERE in the diffuse ISM is consistent with photoluminescence with an
photon efficiency of 10 \Sigma 3% (lower limit) which corresponds to a 4 \Sigma 1% energy efficiency.
We plan to expand on this work by expanding the fraction of the sky studied. The APS
Catalog is scheduled for completion by the end of 1997 (Humphreys 1997) and will cover ¸50% of
the sky. We anticipate using the SKY model (Cohen 1994, 1995) to derive the faint star counts in
regions not covered by the APS catalog after calibrating the SKY model in those regions where
the APS catalog exists. This will allow us to investigate the distribution of ERE and DGL over
almost the entire sky. The all­sky distribution of ERE will allow us to determine the global
ERE photon efficiency as well as any variations as a function of position in the sky. The all­sky
distribution of the DGL will allow us to quantify the variations in the WP model inputs (dust
grain albedo and g, hNHI =E(B \Gamma V )i, and radiation field) on a Galaxy­wide scale. In addition, the
comparison between the SKY model and the APS Catalog will yield valuable information about
Galactic structure.
This paper was part of K. D. Gordon's PhD thesis and, so, thanks to the thesis committee
members Song Cheng, Al Compaan, Steve Federman, Nancy Morrison, Gary Toller, and Adolf
Witt, especially the last three. Thanks to Gary Toller and Jerry Weinberg for their help
understanding how to use the Pioneer measurements. Additional thanks to Gary Toller for
providing the DIRBE maps. Thanks to Chris Cornuelle, Jeff Larson, and Roberta Humphreys
for providing lots of help and information about the APS Catalog of the POSS I. The Pioneer
10 and 11 IPP data were provided by J. L. Weinberg (Principal Investigator) and the National
Space Science Data Center (NSSDC). This research has made use of the APS Catalog of the
POSS I, which is supported by the National Science Foundation, the National Aeronautics and
Space Administration, and the University of Minnesota. The APS database can be accessed at
http://isis.spa.umn.edu/. This research was financially supported under NASA LTSAP grants
NAGW­3168 & NAG5­3367 to The University of Toledo. The COBE datasets were developed by
the NASA Goddard Space Flight Center under the guidance of the COBE Science Working Group
and were provided by NSSDC.
REFERENCES
Arendt, R. G., et al. 1997, in preparation
Aumann, H. H., Fowler, J. W., & Melnyk, M. 1990, AJ, 99, 1674

-- 39 --
Blanco, V. M., Demers, S., Douglas, G. G., & Fitzgerald, M. P. 1968, Publ. U. S. Obs. 2nd Series,
21
Boggess, N. W., et al. 1992, ApJ, 397, 420
Boulanger, F., et al. 1996, A&A, 312, 256
Calzetti, D., Bohlin, R. C., Gordon, K. D., Witt, A. N., & Bianchi, L. 1995, ApJ, 466, L97
Cannon, A. J. 1925­1936, The Henry Draper Extension, Ann. Astron. Obs. Harvard College, 100
Cannon, A. J., & Pickering, E. C. 1918­1924, The Henry Draper Catalogue, Ann. Astron. Obs.
Harvard College, 91­99
Cannon, A. J., & Walton Mayall, M. 1949, The Henry Draper Extension, The Annie J. Cannon
Memorial Volume, Ann. Astron. Obs. Harvard College, 112
Cardelli, J. A., Clayton, G. C., & Mathis, J. S. 1989, ApJ, 345, 245
Chlewicki, G., & Laureijs, R. J. 1987, in Polycyclic Hydrocarbons and Astrophysics, eds. A. L'eger
et al. (Dordrecht: D. Reidel Publishing Co.) 335
Cleary, M. N., Haslam, C. G., & Heiles, C. 1979, A&AS, 36, 95
Cohen, M. 1994, AJ, 107, 582
Cohen, M. 1995, ApJ, 444, 874
Cohen, M. 1997, private communication
Darbon, S., Perrin, J.­M., Sivan, J.­P. 1998, A&A, in press
Diplas, A., & Savage, B. D. 1994, ApJ, 427, 274
Duley, W. W. 1985, MNRAS, 215, 259
Duley, W. W., & Whittet, D.. C. B. 1990, MNRAS, 242, 40P
Duley, W. W., & Williams, D. A. 1988, MNRAS, 230, 1P
Duley, W. W., & Williams, D. A. 1990, MNRAS, 247, 647
Dwek, E., et al. 1997, ApJ, 475, 565
Elmegreen, B. G., & Falgarone, E. 1996, ApJ, 471, 816
Fitzgerald, M. P., Stephens, T. C., & Witt, A. N. 1976, ApJ, 208, 709
Friedmann, B. C. 1996, MS Thesis, The Univ. of Toledo

-- 40 --
Furton, D. G. 1994, PhD Thesis, The Univ. of Toledo
Furton, D. G., & Witt, A. N. 1990, ApJ, 364, L45
Furton, D. G., & Witt, A. N. 1992, ApJ, 386, 587
Furton, D. G., & Witt, A. N. 1993, ApJ, 415, L51
Gordon, K. D., Witt, A. N., Carruthers, G. R., Christensen, S. A., & Dohne, B. C. 1994, ApJ,
432, 641
Gordon, K. D., Witt, A. N., Friedmann, B. C., Schild, R., Bohlin, R. C., & Stecher, T. P. 1997, in
preparation
Greenstein, J. L., & Oke, J. B. 1977, PASP, 89, 131
Guhathakurta, P., & Cutri, R. M. 1994, in The First Symposium on the Infrared Cirrus and
Diffuse Interstellar Clouds, eds. R. M. Cutri & W. B. Latter (San Francisco: BookCrafters,
Inc.) 34
Guhathakurta, P., & Tyson, J. A. 1989, ApJ, 346, 773
Hanner, M. S., Weinberg, J. L., DeShields II, L. M., Green, B. A., & Toller, G. N. 1973,
J. Geophys. Res., 79, 3671
Hauser, M. G., et al. 1997, COBE Diffuse Infrared Background Experiment (DIRBE) Explanatory
Supplement, ed. M. G. Hauser, T. Kelsall, D. Leisawitz, & J. Weiland, COBE Ref. Pub.
No. 95­A (Greenbelt, MD: NASA/GSFC), available in electronic form from the NSSDC
Hayes, D. S., & Latham D. W. 1975, ApJ, 197, 593
Heiles, C., & Cleary, M. N. 1979, Austral. J. Phys., Astroph. Suppl. No. 47, 1
Hoffleit, D., & Warren, W. H. Jr. 1991, The Bright Star Catalogue, 5th Revised Ed. (Preliminary
Version), Astronomical Data Center
Houk, N. 1978, Michigan Catalogue of Two­Dimensional Spectral Types for the HD Stars: vol. 2
(Ann Arbor: Univ. of Michigan)
Houk, N. 1982, Michigan Catalogue of Two­Dimensional Spectral Types for the HD Stars: vol. 3
(Ann Arbor: Univ. of Michigan)
Houk, N., & Cowley, A. P. 1975, University of Michigan Catalogue of Two­Dimensional Spectral
Types for the HD Stars: vol. 1 (Ann Arbor: Univ. of Michigan)
Houk, N., & Smith­Moore, M 1988, Michigan Catalogue of Two­Dimensional Spectral Types for
the HD Stars: vol. 4 (Ann Arbor: Univ. of Michigan)

-- 41 --
Humphreys, R. M., Landau, R., Ghigo, F. D., Zumach, W., & LaBonte, A. E. 1991, AJ, 102, 395
Humphreys, R. M. 1997, private communication
Jenkner, H., Lasker, B. M., Sturch, C. R., McLean, B. J., Shara, M. M., & Russell, J. L. 1990, AJ,
99, 2082
Jenniskens, P. 1994, A&A, 284, 227
Kim, S.­H. & Martin, P. G. 1996, ApJ, 462, 296
Lamla, E. 1982, in Landolt­B¨ornstein, vol. 2, eds. K. Schaifers & H. H. Voigt (Berlin: Springer),
48.
Lasker, B. M., Sturch, C. R., McLean, B. J., Russell, J. L., Jenkner, H., & Shara, M. M. 1990, AJ,
99, 2019
Lemaire, J. L., Field, D., Gerin, M., Leach, S., Pineau des For“ots, G., Rostas, F., & Rouan, D.
1996, A&A, 308, 895
Li, A., & Greenberg, J. M. 1997, A&A, in press
Lockwood, G. W., & T¨ug, H., & White, N. M. 1992, ApJ, 390, 669
Mathis, J. S. 1996, ApJ, 472, 643
Mathis, J. S. 1997, ApJ, submitted
Mathis, J. S., Mezger, P. G., Panagia, N. 1983, A&A, 128, 212
Mattila, K. 1979, A&A, 78, 253
Mattila, K. 1980a, A&AS, 39, 53
Mattila, K. 1980b, A&A, 82, 373
Mermilliod, J.­C. 1987, A&AS, 71, 413
Mermilliod, J.­C. 1994, Bull. Inf. Centre Donn'ees Stellaires, 45, 3
Minkowsi, R. L., & Abell, G. O. 1963, in Stars and Stellar Systems, vol. 2, Basic Astonomical
Data, ed. K. Aa. Strand (Chicago: Univ. of Chicago Press) 481
Ochsenbein, F. 1974, A&AS, 15, 215
Ochsenbein, F., Bischoff, M., & Egret, D. 1981, A&AS, 43, 259
Ochsenbein, F. 1983, Bull. Inf. Centre Donn'ees Stellaires, 24, 109
Odewahn, S. C., Stockwell, E. B., Humphreys, R. M., & Zumach, W. A. 1992, AJ, 103, 318

-- 42 --
Odewahn, S. C., Humphreys, R. M., Aldering, G., & Thurmes, P. 1993, PASP, 105, 1354
Papoular, R., Conard, J., Guillois, O., Nenner, I., Reynaud, C., & Rouzaud, J.­N. 1996, A&A,
315, 222
Pellicori, S. F., Russell, E. E., & Watts, L. A. 1973, Appl. Opt., 12, 1246
Pennington, R. L., Humphreys, R. M., Odewahn, S. C., Zumach, W., & Thurmes, P. M. 1993,
PASP, 105, 521
Perrin, J.­M., Darbon, S., & Sivan, J.­P. 1995, A&A, 304, L21
Perrin, J.­M., & Sivan, J.­P. 1992, A&A, 255, 271
Reynolds, R. J. 1984, ApJ, 282, 191
Rogers, C., Heyer, M. H., & Dewdney, P. E. 1995, ApJ, 442, 694
Roman, N. G., & Warren, W. H. Jr. 1985, Documentation for the Machine­readable Version of the
Henry Draper Catalogue (Edition 1985) NSSDC/WDC­A­R&S 84­18
Russell, J. L., Lasker, B. M., McLean, B. J., Sturch, C. R., & Jenkner, H. 1990, AJ, 99, 2059
Sakata, A., Wada, S., Narisawa, T., Asano, Y., Iijima, Y., Onaka, T., & Tokunaga, A. T. 1992,
ApJ, 393, L83
Smithsonian Astrophysical Observatory Staff 1966, Star Catalog. Positions and Proper Motions of
258,977 Stars for the Epoch and Equinox of 1950.0, Pub. of the Smithsonian Institution of
Washington, D.C. No. 4652 (Washington: Smithonian Institution)
Schmidt, G. D., Cohen, M., & Margon, B. 1980, ApJ, 239, L133
Schmidt­Kaler, Th. 1982, in Landolt­B¨ornstein, vol. 2, eds. K. Schaifers & H. H. Voigt (Berlin:
Springer), 17.
Schuerman, D. W., Giovane, F., & Weinberg, J. L., unpublished; personal communication, J. L.
Weinberg
Scott, A. D., & Evans, A., & Rawlings, J. M. C. 1994, MNRAS, 269, L21
Silva, D. R., & Cornell, M. E. 1992, ApJS, 81, 865
Sivan, J.­P., & Perrin, J.­M. 1993, ApJ, 404, 258
Smartt, S. J., & Rolleston, W. R. J. 1997, ApJ, 481, L47
Snow, T. P., & Witt, A. N. 1995, Science, 270, 1455
Snow, T. P., & Witt, A. N. 1996, ApJ, 468, L65

-- 43 --
Sofia, U. J., Cardelli, J. A., Guerin, K. P., & Meyer, D. M. 1997, ApJ, 482, L105
Stark, A. A., Gammie, C. F., Wilson, R. W., Bally, J., Linke, R. A., Heiles, C., & Hurwitz, M.
1992, ApJS, 79, 77
Szomoru, A. & Guhathakurta, P. 1998, ApJ, in press
Toller, G. N. 1981, PhD Thesis, State Univ. of New York at Stony Brook
Toller, G. N. 1997, private communication
Toller, G. N., Tanabe, H., Weinberg, J. L. 1987, A&A, 188, 24
T¨ug, H., White, N. M., & Lockwood, G. W. 1977, A&A, 61, 679
van Breda, I. G., & Whittet, D. C. B. 1981, MNRAS, 195, 79
Watkin, S., Geldhill, T. M., & Scarrott, S. M. 1991, MNRAS, 1991, 252, 229
Webster, A. 1993, MNRAS, 264, L1
Weinberg, J. L., Hanner, M. S., Beeson, D. E., DeShields II, L. M., & Green, B. A. 1974,
J. Geophys. Res., 79, 3665
Weinberg, J. L., & Schuerman, D. W. 1981, User's Guide for Pioneer 10/11 Background Sky Tape
Whittet, D. C. B. 1992, Dust in the Galactic Environment (Bristol: IOP)
Wicenec, A., & van Leeuwen, F. 1995, A&A, 304, 160
Witt, A. N. 1988, in Dust in the Universe, eds. M. E. Bailey & D. A. Williams (Cambridge:
Cambridge Univ. Press) 1
Witt, A. N., & Boroson, T. A. 1990, ApJ, 355, 182
Witt, A. N., Friedmann, B. C., Sasseen, T. P. 1997, ApJ, 481, 809
Witt, A. N., & Furton, D. G. 1995, in The Diffuse Interstellar Bands, eds. A. G. G. M. Tielens &
T. P. Snow (Dordrecht: Kluwer) 149
Witt, A. N. & Gordon, K. D. 1996, ApJ, 463, 681
Witt, A. N., & Johnson, M. W. 1973, ApJ, 181, 363
Witt, A. N., & Malin, D. F. 1989, ApJ, 347, L25
Witt, A. N., Oliveri, M. V., & Schild, R. E. 1990, AJ, 99, 888

-- 44 --
Witt, A. N., & Petersohn, L. K. 1994, in ASP Conf. Ser. 58, The First Symposium on the Infrared
Cirrus, and Diffuse Interstellar Clouds, eds. R. M. Cutri & W. B. Latter (San Francisco:
ASP)
Witt, A. N., & Schild, R. E. 1985, ApJ, 294, 225
Witt, A. N., & Schild, R. E. 1986, ApJS, 62, 839
Witt, A. N., & Schild, R. E. 1988, ApJ, 325, 837
Witt, A. N., Schild, R. E., & Kraiman, J. B. 1984, ApJ, 281, 708
Witt, A. N., Stecher, T. P., Boroson, T. A., Bohlin, R. C. 1989b, ApJ, 336, L21
Witt, A. N., Walker, G. A. H., Bohlin, R. C., & Stecher, T. P. 1982, ApJ, 261, 492
Zubko, V. G., Kre/lowski, J., & Wegner, W. 1996, MNRAS, 283, 577
This preprint was prepared with the AAS L A T E X macros v4.0.